Mitchell1 Revs Up Results With NoSQL
How to Increase the Reliability of Your IT Infrastructure Using Predictive Analytics REGISTER >
A leading content publisher for the automotive industry turns to a NoSQL database to improve search functionality and more.
Over the last half century, automobiles have become far more complex, and manuals and other documents have grown fatter and more challenging to use. In many cases, technicians must sort through countless pages of documents or hundreds of hits after a search in order to find the relevant information.
"Finding the right repair information can become an incredibly difficult and time-consuming task," said Jeff Grier, senior director for product management at Mitchell1, a firm that supplies online OEM and aftermarket repair information to dealers and repair shops throughout North America and beyond.
As a result, Mitchel1, which licenses content from 27 different companies and has more than 77,000 customers, has continued to roll into the digital age. In February 2012, it introduced ProDemand, a new product that uses the MarkLogic Enterprise NoSQL database platform to pull together disparate data sources, including manuals, recall orders and other documents and minimize data complexity for end users. Technicians access the information through dedicated computing devices in their shops as well as iPads and Android tablets. The information is typically pushed out through Mitchell1's ProDemand software, which works with Web browsers. "In the past, repair technicians often found themselves drowning in too many results that were not focused appropriately. If they searched on 'fuel pump,' for example, they would encounter numerous mentions and the results often weren't weighted effectively," Grier said.
Making things even more challenging, Mitchell1 typically updates and releases content every two weeks--along with regularly adding features. "We have an editorial team constantly working on the metadata and tags to ensure that they are current and accurate." Prior to the NoSQL database conversion, the task was becoming increasingly difficult. "The previous data structure simply wasn't flexible enough to keep up." By contrast, the NoSQL database puts words and data into context and provides a highly flexible and agile framework for managing tasks.
For example, in fall of 2014, the company added a feature that allows it to show the relationship between codes that are attached to engines and components. This helps improve search results. "It tells technicians where to start their research by displaying the most common or likely repairs for the specific situation," he explained. "The system uses information from hundreds of millions of repair orders about a specific make, model and year to spot patterns and show that, say, 85 percent of the time the technician should take a look at a certain part or issue first." Another addition revolved around adapting vocabulary to fit regional language differences in the U.S. across 275,000 automotive terms.
The results of the database initiative have helped Mitchell1 accelerate value to customers and improve bottom line results. Search tasks that previously took as long as 20-30 minutes now sometimes take place in seconds. In fact, Grier said that the previous system typically required 15 times longer for a technician to get to useful information. Internally, it's also allowing teams to focus on more strategic tasks and create further value for customers. "There is an order of magnitude improvement across the board," he explained.